For a well-maintained ship, significant fractures are caused by one or more of the following;16 Abnormal forces in or on the ship structure; Presence of flaws or notches in the structure where fractures originate; and Inadequate physical properties of the structural steel at service temperatures. All three factors were involved in the side shell fracture of the LakeCarling. Minor cracks, as opposed to significant fractures, are a fact of life on bulk carriers, or any type of large vessel, for that matter. Of major importance is the vessel's damage tolerance, that is to say, the length to which a through thickness flaw or crack can grow before becoming critical. Assuming that loading and dynamic forces remain within design parameters, the fracture toughness of the metal is what will ultimately determine this length.Analysis For a well-maintained ship, significant fractures are caused by one or more of the following;16 Abnormal forces in or on the ship structure; Presence of flaws or notches in the structure where fractures originate; and Inadequate physical properties of the structural steel at service temperatures. All three factors were involved in the side shell fracture of the LakeCarling. Minor cracks, as opposed to significant fractures, are a fact of life on bulk carriers, or any type of large vessel, for that matter. Of major importance is the vessel's damage tolerance, that is to say, the length to which a through thickness flaw or crack can grow before becoming critical. Assuming that loading and dynamic forces remain within design parameters, the fracture toughness of the metal is what will ultimately determine this length. Side Shell Crack Initiation The LakeCarling underwent a survey in dry dock approximately one year before the occurrence. Particular attention had been paid to the bottoms of the side shell frames due to the renewal of many of them, including frame91 port. Since none of the six crack locations in No.4hold had been previously repaired, and the four in No.2hold had been repaired only superficially, it is highly unlikely that these cracks were present at the time of the dry docking. Crack initiation may be due to any number of causes, including: improper deballasting during loading; insufficient draughts while transiting a seaway in ballast; asymmetrical loading; damage by unloading grabs during discharge; side shell striking while negotiating locks; or exceeding the approved seagoing SWBM. Unsatisfactory welding procedures and localized construction details can also cause or contribute to the initiation of such cracks. Once initiated, cracks will, depending on the operational environment of the vessel, usually enter a stage of slow, stable growth. Superficially, similarities seem to exist amongst all ten crack locations, thus implying that they were created by the same mechanism, but some major differences are also evident between those in No.4hold and those in No.2hold. In No.4hold, all six of the crack locations, three to port and three to starboard, are concentrated in an area within eight frames, and at each location the cracks are roughly symmetrical fore and aft of the frame. All were unrepaired and three of the frames had been replaced during the Gdansk dry dock. In contrast, at the four crack locations in No.2hold, the cracks are not all symmetrical fore and aft of the frame, and in the case of frame171, there is no crack aft of the frame. Furthermore, they are not concentrated in a limited area of the hold; two are in the forward section and two in the aft section of the hold. All four of the cracks in No.2hold were superficially repaired and only one frame had been replaced at the Gdansk drydock. Another major difference is one of construction detail. All of the frames of concern in No.4hold are of the separate bracket configuration while those in No.2hold are integral brackets. The stress concentration factors, such as the discontinuities caused by the scallop (cut-out) in the side frame and the proximity to the change in plate thickness at the shell plate seam weld, are not entirely similar. Of the ten crack locations, four of the frames had been cropped and renewed at the bottom. There does not seem to be a strong correlation between frames replaced in dry dock and the crack locations, but the correlation cannot be discounted entirely. Half of the crack locations (3of6) in No.4hold were where frames had been cropped and renewed. Given the preceding, it is most likely that the cracks in No.4hold were created by the same mechanism at some time between the dry dock in Gdansk and the loading at Sept-les. Although the cracks in No.2hold were probably created during this same time frame, it is less certain that they were created by the same mechanism as those in No.4hold. Several sources could have been responsible for the cracks in No.4hold. De-ballasting in unprotected waters, and/or the improper loading four months prior to the hull fracture, are possible causes of the crack initiation. For the de-ballasting scenario, the SWBM imposed on the hull girder at frame91 would have been 107%of the approved maximum permissible. For the loading scenario, the SWBM at frame86 was 103%of the approved seagoing allowable limit. Being farthest from the neutral axis maximum stresses would have been experienced in the deck and bottom shell. However, the combination of all global and localized stresses would still have been significant at the bottom of the side shell frames. The vessel sailed in this condition for 5days, from ThunderBay to Montral, in water close to 5C. After leaving Montral, the vessel encountered very heavy weather in the North Atlantic. Had small cracks developed due to improper loading and cold water conditions between Thunder Bay and Montral, they could have grown under such dynamic loading. The restrained nature of the welded connections at the lower ends of the side shell frames made this area susceptible to the retention of residual stresses. The coincidence of several stress concentration factors, such as: the discontinuities caused by the scallop (cut-out) in the side frame; the proximity of the frames lower ends to the shell plate seam (possibly exacerbated when the frames were renewed at Gdansk); the change in plate thickness at the shell plate seam weld; and the presence of residual stresses; created the conditions necessary, when subjected to high stresses and cold ambient temperatures, to cause small cracks to form at the base of the side shell frames between frames85 and96 in No.4hold.17 The intervening four months operation prior to the occurrence is a reasonable time frame in which these cracks could grow imperceptibly under the dynamic loading of the hull girder. Side Shell Fracture Properly loaded at Sept-les and in relatively calm seas, no relationship can be drawn between the fracture and these operational and environmental factors. Ultimately, the small crack at frame91 went critical solely due to factors related to the physical properties of the steel and the ambient temperature. The grade A steel used in the construction of the side shell of the LakeCarling was within specifications insofar as tensile strength is concerned, but as for minimum CVN, no specifications actually exist. The relatively low fracture toughness of the side shell plate when exposed to temperatures near 0C allowed the forward crack at frame91 (port) to grow to failure at a load well below the ultimate tensile strength of the material. The length of this crack at the time it became critical was not determined but calculations have shown it could have been as short as 10cm. According to the IACS Unified Rules, grade A steel less than 50mm thick (and gradeB 25mm or less in thickness) does not have to demonstrate a minimum CVN. Under these rules this steel can be used for a ship's side shell. Some testing has shown that the average CVN of grade A steel available worldwide is often quite high and the grain size relatively small.18 This, in effect, sets a defacto standard - ship owners, ship constructors, and classification societies all expect and depend upon grade A steel having a fracture toughness that is sufficient for all operational conditions. However, without actual standards, expectations are not always enough to ensure adequate fracture toughness and damage tolerance. Although the relationship between CVN energy and fracture toughness is not necessarily straightforward, the system has been used with relative success by all of the major classification societies for many years by providing a qualitative estimate of material toughness. There are, however, no requirements to use steel of a given CVN energy at low operating temperatures in way of the ship's sides (which are usually gradeA steel). Nonetheless, cargo vessels may often trade in zones where ambient temperatures are close to, or below, 0C and these low temperatures generally tend to reduce the ability of the steel to resist crack growth. The grain size and CVN impact energy, and thus the corresponding fracture toughness, of the gradeA steel used in the construction of the LakeCarling was well below the defacto standard when compared to average values of grade A steels available worldwide. This steel performed below expectations and did not provide a reasonable damage tolerance in all operational conditions. Steel Toughness Standards and Damage Tolerance The LakeCarling was relatively new and had been recently inspected, yet a substantial fracture resulted from the existence of what should have been a tolerable crack (10cm) in the ship's side. The LakeCarling occurrence, although seemingly rare, is most certainly not unique. Historical data have revealed that nearly three quarters of all casualty-related fatalities on bulk carriers are attributable to structural failure.19 Other data culled from Lloyd's casualty database indicate 23bulk carriers foundered in cold water in a twenty-year period, yet the cause of the losses are undetermined.20 Notable vessel losses in the TSB databank are as follows: Jalamorari, General Cargo, December1982. Charlie, Bulk Carrier, January1990. Protektor, Bulk Carrier, January1991. Marika, Bulk Carrier, January1994. SalvadoreAllende, December1994. LeaderL, March 2000. Albeit almost always in heavy weather, these losses were also all in cold temperatures. Due to a lack of forensic evidence, the true cause of these losses cannot be proven. Although the Enhanced Survey Program (ESP) and other initiatives more recently introduced to reduce risk for bulk carriers are continuing to increase safety, the LakeCarling can be viewed as an example of residual risk that remains in spite of these initiatives. A recent evaluation by IACS of risk control options (RCO) in respect of the side shell integrity of bulk carriers identified 15RCOs, 11of which were put forward for further investigation.21 Although one option called for the requirement to use notch toughened steel and associated welding consumables for frame brackets, toughness of the metal used in the side shell was not addressed or identified as a RCO. The appropriateness of using steel of unknown toughness in vessel construction has been raised in various reports and proceedings, including those concerning the loss of the Derbyshire, the brittle fractures of the TyneBridgeand the breaking in two of the Kurdistan.22 During the re-opened Derbyshire inquiry (under Justice Coleman ([U.K.]), the following quotation was restated: Depending on the properties of the steel float and/or weld, the ambient temperature and the location of the crack, a crack as small as 30millimetres could be sufficient to initiate a fast-running brittle fracture.23 The steel toughness of the Derbyshire was not further investigated because no steel was actually taken from the wreck for testing. In his independent analysis of the Derbyshire sinking, the Professor Emeritus of Naval Architecture at the University of Glasgow, Scotland, D. Faulkner, stated his support for reviewing the use of metal of unknown fracture toughness in ship's hulls.24 Although the recent Lloyd's initiative to qualify the toughness of grade A steel may appear to be an improvement on existing standards, the required 27Joules at 20C is less than that demonstrated by the LakeCarling; and 20C is certainly well above the temperature most vessels may expect to encounter at one time or another. Additionally, Lloyd's leaves it up to the manufacturer to report that the steel meets this requirement by way of in-house checks. This measure, although well intentioned, is less a tool for quality control than it is an indication that the toughness of gradeA steel has been, and continues to be a cause for concern. It has been suggested that a FATT below 0C is necessary to ensure sufficient fracture toughness for ship's hulls.25 In the Lloyd's study of the fracture properties of gradeA steel, 5of 39samples (nearly13%) demonstrated a FATT above 0C, while a further four samples (10%) were at -6C or above. For the LakeCarling, the FATT was determined to be 32C. In other industries, such as electric power generation, risks due to brittle fracture are reduced by ensuring that operating pressures are only permitted at component temperatures approaching or exceeding the component's FATT.26 A recent study found, after a review of the available data, a significant variability in the fracture initiation toughness of gradeA plates.27 Other studies have found similar results and have advocated the use of a prescribed minimum toughness standard for all metal and welds used in ship hulls.28 In fact, 40J at -40C has been the standard for Canadian ships of war for over 40years, while 100J at -20C has also been suggested as a minimum to ensure adequate damage tolerance and protection against brittle fracture.29 In a major review of a vast amount of available literature concerning the fracture properties of grade A ship plate, it was concluded that ...the crack arrest ability of grade A plate is poor and probably inadequate for most ship applications.30 Nonetheless, it would appear that, notwithstanding the average high toughness and quality of most steels, some gradeA andB steels that are not suitable in all conditions are still being produced and used in ship's hulls. In the marine industry, standards evolve over time, usually in reaction to a high profile disaster or event. Because of the nature of the trade, bulk carriers are prone to side shell flexing, and the side shell is more at risk from crack damage than any other area of the vessel.31 When ships are lost without a trace or are inaccessible, it is not possible to analyse the relationship between material toughness and the cause of the vessel's loss. The LakeCarling had loaded iron ore in the port of Sept-les and the fracture was discovered when the vessel was close offshore in the Gulf of St. Lawrence. This provided the TSB with an opportunity to closely examine the fractures and conduct an in-depth analysis of all aspects of the occurrence, including: the circumstances leading to the occurrence, the cause of the fracture and the inherent mechanical properties of the steel. One certainty remains - all ships, especially bulk carriers, operating in cold waters and having their side shell of metal with characteristics similar to those of the LakeCarling, are at risk. The damage tolerance could be less than adequate and cracks could remain unnoticed or discounted as insignificant, yet they would still pose a significant risk when exposed to low temperatures. Given the uncertainties and variability of fracture toughness for some gradeA andB steels, it would appear that residual risks for unstable brittle fracture are still present in vessels with hulls constructed with these steels, especially when operating in colder climates. Unreported repairs The cracks in No.2hold were repaired in a substandard fashion and were not reported to the classification society. In its report into the structural failure and sinking of the bulk carrier LeaderL, the Polish Classification Society concluded: To assure its local strength, the structure should also be continuously supervised. This requires close co-operation of the classification society, shipowner and crew (to record noticed damages and defects), which not always is the case.32 One of the major risk reduction measures implemented in the1990s addressing structural failures in bulk carriers has been the ESP. It has been shown in one study that the ESP has had a general effectiveness in the order of 19% for these vessels within this category of casualty.33 Notwithstanding being under the ESP regime, some cracks in the LakeCarling hull went unnoticed and unrepaired. Those cracks that were repaired were not executed to classification society specifications nor were they reported. This omission increased risks to the vessel and crew. Immersion Suits Although the LakeCarling was carrying the required minimum number of immersion suits - one for each member of the rescue boat crew - in the first hours that followed the discovery of the fracture the master requested that additional suits be dropped by SAR aircraft. This was a prudent decision even though, in the end, they were not used. Because SAR resources and the extra suits were readily available, the drop was possible. Since their introduction into the marine industry, immersion suits have proven to be an efficient and reliable defence against death by hypothermia. On Canadian vessels the carriage of immersion suits for all crew members has been mandatory since1983.34 The TSB has recorded numerous instances where immersion suits have saved lives: December 1990, a crew member of a fishing vessel rescued after seven hours in cold water; January 1993, a crew member of a fishing vessel was recovered after approximately five hours in the frigid sea; February 1995, a crew member of a fishing vessel was rescued after over two hours in cold water; December 2001, of a four man crew, both persons wearing immersion suits survived while, of the other two (not wearing immersion suits), only one survived. In 2001, subsequent to the Flare investigation, Canadasubmitted a proposal to the 74thsession of the IMO Maritime Safety Committee (MSC).35 In2002, the MSC Sub-Committee on Ship Design and Equipment (DE) considered the carriage of immersion suits for all persons on board cargo .vessels should be made mandatory, particularly in cases where casualties occurred in cold climates. In certain circumstances, individuals involved may then have a better chance for survival and rescue. The DE Sub-Committee meeting in March2003 further considered the issue and subject to, interalia, a geographical definition of warm climates where carriage of immersion suits would not be required, developed and submitted to MSC a draft of proposed amendments to SOLAS regulationIII/32.3, (Personal Life-saving Appliances). The TSB has found that some residual risks appear to remain, even when carrying immersion suits for 100%of the crew, particularly with respect to the maintenance of the zippers. Past investigations have shown that poor zipper maintenance can nullify the advantages of having an immersion suit. Hand in hand with any new requirements for more widespread carriage of immersion suits should be provisions for training and proper maintenance of this equipment. The DE Sub-Committee is presently developing guidelines for periodic testing of immersion (and anti-exposure) suit seams and closures for consideration by MSC. The TSB commends these initiatives. Conclusions Findings as to Causes and Contributing Factors service loads greater than those approved for the vessel; probable presence of residual stress; stress concentration factors due to discontinuity caused by scallop (cut-out) in the side frame; the proximity of the frame end to the shell plate seam weld; and the change in plate thickness at the shell plate. service loads greater than those approved for the vessel; probable presence of residual stress; stress concentration factors due to discontinuity caused by scallop (cut-out) in the side frame; the proximity of the frame end to the shell plate seam weld; and the change in plate thickness at the shell plate. The relatively low fracture toughness of the side shell plate when exposed to near 0C temperatures allowed the forward crack at frame91(port) to grow to failure at a load well below the ultimate tensile strength of the material. The length of this crack at the time it became critical was not determined but could have been as short as 10cm. Approximately four months before this occurrence, the LakeCarling was subjected to service loads that exceeded the maximum approved seagoing bending moment. Findings as to Risk There are no Unified Requirements to use steel of a certified toughness or minimum FATT in way of the ship's sides for cargo vessels which may often trade in zones where ambient temperatures are close to, or below, 0C. Given the variability and unqualified fracture toughness for some gradeA and Bsteels, it would appear that residual risks for unstable brittle fractures are present in vessels with hulls constructed with these steels, especially when operating in colder climates. The large grain size and low CVN impact energy of the LakeCarling's side shell plate resulted in a corresponding fracture toughness that is below expectations and does not permit a reasonable damage tolerance in all operational conditions. Cracks at the bases of four side frames in No.2hold had been observed and repairs had been made. These cracks and subsequent repairs were not documented or reported to the Classification society, nor were they completed in accordance to the Classification society's specifications. The LakeCarling complied with SOLAS minimum requirements for the carriage of immersion suits. However, although the vessel often operated in areas of sub- zero weather, immersion suits were not carried for all crew members - nor are they currently required to be carried. Several side shell frames were repaired in Gdansk a year before the side shell failure. Although there does not appear to be a strong correlation between the principal fracture (and other cracks discovered at the base of the frames) and these repairs, it cannot be discounted entirely. Other Findings Although built to specifications that allowed alternate cargo hold loading, the LakeCarling was rarely loaded in this manner. Greater SWBMs are imposed on the structure when alternate hold loading is adopted. Safety Action Action Taken Although not specifically related to events of the LakeCarling fracture, discussions at IMO have addressed alternate hold loading; specifically the possible benefits deriving from banning alternate hold loading of heavy cargoes in the full load condition, and in particular the resulting reduction in shear forces and bending moments when loading homogeneously in all holds.36 Further meetings of the Maritime Safety Committee (MSC) agreed that the Design and Equipment (DE) sub-committee develop draft amendments to SOLAS chapter XII along the following lines: Bulk carriers in the full load condition (90%of the ship's deadweight at the relevant freeboard) of single-side skin construction and 150m in length and over, constructed before 1July1999, after reaching 10years of age, or constructed after 1July1999 if not in compliance with SOLAS chapterXII and IACSURS12Rev2.1, shall be banned from sailing with any hold empty. The ban shall not apply to ships constructed before 1July1999 if they comply with SOLAS chapterXII and IACSURS12Rev2.1.37 The proposal will be further discussed at the 2004DE47 sub-committee meeting. Safety Concern The use of grade A and grade B steel of unknown toughness or fracture appearance transition temperature (FATT) in way of ships' side shells has, in the past and to this day, allowed some vessels to be constructed of steel that is less than adequate for all ambient conditions. Because a vessel's side shell, particularly bulk carriers, is prone to flexing, the side shell is more at risk to crack damage than any other area of the vessel. Crack initiation is the first step towards a major fracture. Once a crack has initiated, only the material's damage tolerance stands between a nuisance defect and disaster. The material's damage tolerance is intimately related to its inherent toughness - a quality that can change dramatically for the worse in temperatures at or near 0C if certain characteristics of the steel, such as carbon content or grain size are less than optimal. Over the past 50years, the debate amongst and between the various Classification Societies and other materials experts has been divided. On the one hand, the status quo is touted as sufficient and ample defence against brittle fracture. The status quo, however, is a moving target. The standards of today are more rigorous than in 1950 - thanks, in no small measure, to some well documented disasters. On the other hand, objective evidence and a review of the pertinent literature has indicated, and eminent world leaders in the field have emphasized, the lack of toughness standards for this aspect of ship construction. In a recent review of statistics over the period 1988-1998, of ships over 500gt, close to 50percent of all the causes of the total loss of a vessel were attributable to either weather or various.38 It is conceivable that somewhere within those statistics are other instances of structural failure. Without doubt, a considerable portion of these losses could be due to structural failure - and many of those structural failures could be attributed to brittle fracture. Because most of the wrecks can not be sufficiently investigated, the causes are attributed to weather or various. However, the use of weather as a cause, although it may have contributed to the occurrence, is not considered appropriate as a criterion in some cases since modern vessels are built to withstand weather. Although the average Charpy V-Notch (CVN) energy of today's grade A and B steel can generally be expected to be relatively high, 33% of the samples tested by Lloyd's had a fracture appearance transition temperature (FATT) greater than -10C. Furthermore, five of the 39samples (12.8%) had a FATT greater than 0C. Any reasonable assessment of these results should conclude the existence of less than adequate toughness. By any definition, even requiring 27J at 20C is a low standard - but it is a standard. The very fact that grade A steel is, by definition, a steel without a toughness standard should raise concerns. Such action as identifying cargo hold water level detectors as a reasonable defence and risk reduction factor is not without merit, but this is a defence that is reactive rather than proactive. The Board is encouraged with the International Association of Classification Societies' (IACS) intention to carry out critical crack length calculations taking into account the actual material characteristics included in this report. Based on the results of this analysis, IACS will apparently consider whether (or not) to introduce a screening of the material properties of shell plating in way of the single skin areas of the cargo and machinery region in ships with ice strengthening. The Board is also encouraged with the work of IMO involving restrictions on alternate hold loading and their proposal for Goal-based new ship construction standards. The Board is concerned, however, that even if a standard is agreed upon, too low a standard would cause unwanted and necessary constraints with a questionable safety benefit. Furthermore, until such time that restrictions or regulations are put into effect, existing bulk carriers and their crews continue to be at risk. Additionally, even vessels without ice strengthening are regularly called upon to trade in waters with sea temperatures at or near 0C. By limiting any possible modifications of the IACS UR S6 (Use of steel grades for various hull members) to ice-strengthened vessels, other vessels would continue to be exposed to unacceptable residual risks. The Board will continue to monitor this safety issue.